[libav.git] / libavcodec / vdpauvideo.c

/*
 * Video Decode and Presentation API for UNIX (VDPAU) is used for
 * HW decode acceleration for MPEG-1/2, H.264 and VC-1.
 *
 * Copyright (c) 2008 NVIDIA
 *
 * This file is part of FFmpeg.
 *
 * FFmpeg is free software; you can redistribute it and/or
 * modify it under the terms of the GNU Lesser General Public
 * License as published by the Free Software Foundation; either
 * version 2.1 of the License, or (at your option) any later version.
 *
 * FFmpeg is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * Lesser General Public License for more details.
 *
 * You should have received a copy of the GNU Lesser General Public
 * License along with FFmpeg; if not, write to the Free Software
 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
 */

#include <limits.h>
#include "avcodec.h"
#include "h264.h"

#undef NDEBUG
#include <assert.h>

#include "vdpau.h"
#include "vdpau_internal.h"

/**
 * \addtogroup VDPAU_Decoding
 *
 * @{
 */

void ff_vdpau_h264_set_reference_frames(MpegEncContext *s)
{
    H264Context *h = s->avctx->priv_data;
    struct vdpau_render_state * render, * render_ref;
    VdpReferenceFrameH264 * rf, * rf2;
    Picture * pic;
    int i, list, pic_frame_idx;

    render = (struct vdpau_render_state*)s->current_picture_ptr->data[0];
    assert(render);

    rf = &render->info.h264.referenceFrames[0];
#define H264_RF_COUNT FF_ARRAY_ELEMS(render->info.h264.referenceFrames)

    for (list = 0; list < 2; ++list) {
        Picture **lp = list ? h->long_ref : h->short_ref;
        int ls = list ? h->long_ref_count : h->short_ref_count;

        for (i = 0; i < ls; ++i) {
            pic = lp[i];
            if (!pic || !pic->reference)
                continue;
            pic_frame_idx = pic->long_ref ? pic->pic_id : pic->frame_num;

            render_ref = (struct vdpau_render_state*)pic->data[0];
            assert(render_ref);

            rf2 = &render->info.h264.referenceFrames[0];
            while (rf2 != rf) {
                if (
                    (rf2->surface == render_ref->surface)
                    && (rf2->is_long_term == pic->long_ref)
                    && (rf2->frame_idx == pic_frame_idx)
                )
                    break;
                ++rf2;
            }
            if (rf2 != rf) {
                rf2->top_is_reference    |= (pic->reference & PICT_TOP_FIELD)    ? VDP_TRUE : VDP_FALSE;
                rf2->bottom_is_reference |= (pic->reference & PICT_BOTTOM_FIELD) ? VDP_TRUE : VDP_FALSE;
                continue;
            }

            if (rf >= &render->info.h264.referenceFrames[H264_RF_COUNT])
                continue;

            rf->surface             = render_ref->surface;
            rf->is_long_term        = pic->long_ref;
            rf->top_is_reference    = (pic->reference & PICT_TOP_FIELD)    ? VDP_TRUE : VDP_FALSE;
            rf->bottom_is_reference = (pic->reference & PICT_BOTTOM_FIELD) ? VDP_TRUE : VDP_FALSE;
            rf->field_order_cnt[0]  = pic->field_poc[0];
            rf->field_order_cnt[1]  = pic->field_poc[1];
            rf->frame_idx           = pic_frame_idx;

            ++rf;
        }
    }

    for (; rf < &render->info.h264.referenceFrames[H264_RF_COUNT]; ++rf) {
        rf->surface             = VDP_INVALID_HANDLE;
        rf->is_long_term        = 0;
        rf->top_is_reference    = 0;
        rf->bottom_is_reference = 0;
        rf->field_order_cnt[0]  = 0;
        rf->field_order_cnt[1]  = 0;
        rf->frame_idx           = 0;
    }
}

void ff_vdpau_add_data_chunk(MpegEncContext *s,
                             const uint8_t *buf, int buf_size)
{
    struct vdpau_render_state * render;

    render = (struct vdpau_render_state*)s->current_picture_ptr->data[0];
    assert(render);

    render->bitstream_buffers= av_fast_realloc(
        render->bitstream_buffers,
        &render->bitstream_buffers_allocated,
        sizeof(*render->bitstream_buffers)*(render->bitstream_buffers_used + 1)
    );

    render->bitstream_buffers[render->bitstream_buffers_used].struct_version  = VDP_BITSTREAM_BUFFER_VERSION;
    render->bitstream_buffers[render->bitstream_buffers_used].bitstream       = buf;
    render->bitstream_buffers[render->bitstream_buffers_used].bitstream_bytes = buf_size;
    render->bitstream_buffers_used++;
}

void ff_vdpau_h264_picture_complete(MpegEncContext *s)
{
    H264Context *h = s->avctx->priv_data;
    struct vdpau_render_state * render;

    render = (struct vdpau_render_state*)s->current_picture_ptr->data[0];
    assert(render);

    render->info.h264.slice_count = h->slice_num;
    if (render->info.h264.slice_count < 1)
        return;

    for (int i = 0; i < 2; ++i) {
        int foc = s->current_picture_ptr->field_poc[i];
        if (foc == INT_MAX)
            foc = 0;
        render->info.h264.field_order_cnt[i] = foc;
    }

    render->info.h264.is_reference                           = s->current_picture_ptr->reference ? VDP_TRUE : VDP_FALSE;
    render->info.h264.frame_num                              = h->frame_num;
    render->info.h264.field_pic_flag                         = s->picture_structure != PICT_FRAME;
    render->info.h264.bottom_field_flag                      = s->picture_structure == PICT_BOTTOM_FIELD;
    render->info.h264.num_ref_frames                         = h->sps.ref_frame_count;
    render->info.h264.mb_adaptive_frame_field_flag           = h->sps.mb_aff;
    render->info.h264.constrained_intra_pred_flag            = h->pps.constrained_intra_pred;
    render->info.h264.weighted_pred_flag                     = h->pps.weighted_pred;
    render->info.h264.weighted_bipred_idc                    = h->pps.weighted_bipred_idc;
    render->info.h264.frame_mbs_only_flag                    = h->sps.frame_mbs_only_flag;
    render->info.h264.transform_8x8_mode_flag                = h->pps.transform_8x8_mode;
    render->info.h264.chroma_qp_index_offset                 = h->pps.chroma_qp_index_offset[0];
    render->info.h264.second_chroma_qp_index_offset          = h->pps.chroma_qp_index_offset[1];
    render->info.h264.pic_init_qp_minus26                    = h->pps.init_qp - 26;
    render->info.h264.num_ref_idx_l0_active_minus1           = h->pps.ref_count[0] - 1;
    render->info.h264.num_ref_idx_l1_active_minus1           = h->pps.ref_count[1] - 1;
    render->info.h264.log2_max_frame_num_minus4              = h->sps.log2_max_frame_num - 4;
    render->info.h264.pic_order_cnt_type                     = h->sps.poc_type;
    render->info.h264.log2_max_pic_order_cnt_lsb_minus4      = h->sps.log2_max_poc_lsb - 4;
    render->info.h264.delta_pic_order_always_zero_flag       = h->sps.delta_pic_order_always_zero_flag;
    render->info.h264.direct_8x8_inference_flag              = h->sps.direct_8x8_inference_flag;
    render->info.h264.entropy_coding_mode_flag               = h->pps.cabac;
    render->info.h264.pic_order_present_flag                 = h->pps.pic_order_present;
    render->info.h264.deblocking_filter_control_present_flag = h->pps.deblocking_filter_parameters_present;
    render->info.h264.redundant_pic_cnt_present_flag         = h->pps.redundant_pic_cnt_present;
    memcpy(render->info.h264.scaling_lists_4x4, h->pps.scaling_matrix4, sizeof(render->info.h264.scaling_lists_4x4));
    memcpy(render->info.h264.scaling_lists_8x8, h->pps.scaling_matrix8, sizeof(render->info.h264.scaling_lists_8x8));

    ff_draw_horiz_band(s, 0, s->avctx->height);
    render->bitstream_buffers_used = 0;
}

void ff_vdpau_mpeg_picture_complete(MpegEncContext *s, const uint8_t *buf,
                                    int buf_size, int slice_count)
{
    struct vdpau_render_state * render, * last, * next;
    int i;

    render = (struct vdpau_render_state*)s->current_picture_ptr->data[0];
    assert(render);

    /* fill VdpPictureInfoMPEG1Or2 struct */
    render->info.mpeg.picture_structure          = s->picture_structure;
    render->info.mpeg.picture_coding_type        = s->pict_type;
    render->info.mpeg.intra_dc_precision         = s->intra_dc_precision;
    render->info.mpeg.frame_pred_frame_dct       = s->frame_pred_frame_dct;
    render->info.mpeg.concealment_motion_vectors = s->concealment_motion_vectors;
    render->info.mpeg.intra_vlc_format           = s->intra_vlc_format;
    render->info.mpeg.alternate_scan             = s->alternate_scan;
    render->info.mpeg.q_scale_type               = s->q_scale_type;
    render->info.mpeg.top_field_first            = s->top_field_first;
    render->info.mpeg.full_pel_forward_vector    = s->full_pel[0]; // MPEG-1 only.  Set 0 for MPEG-2
    render->info.mpeg.full_pel_backward_vector   = s->full_pel[1]; // MPEG-1 only.  Set 0 for MPEG-2
    render->info.mpeg.f_code[0][0]               = s->mpeg_f_code[0][0]; // For MPEG-1 fill both horiz. & vert.
    render->info.mpeg.f_code[0][1]               = s->mpeg_f_code[0][1];
    render->info.mpeg.f_code[1][0]               = s->mpeg_f_code[1][0];
    render->info.mpeg.f_code[1][1]               = s->mpeg_f_code[1][1];
    for (i = 0; i < 64; ++i) {
        render->info.mpeg.intra_quantizer_matrix[i]     = s->intra_matrix[i];
        render->info.mpeg.non_intra_quantizer_matrix[i] = s->inter_matrix[i];
    }

    render->info.mpeg.forward_reference          = VDP_INVALID_HANDLE;
    render->info.mpeg.backward_reference         = VDP_INVALID_HANDLE;

    switch(s->pict_type){
    case  FF_B_TYPE:
        next = (struct vdpau_render_state*)s->next_picture.data[0];
        assert(next);
        render->info.mpeg.backward_reference     = next->surface;
        // no return here, going to set forward prediction
    case  FF_P_TYPE:
        last = (struct vdpau_render_state*)s->last_picture.data[0];
        if (!last) // FIXME: Does this test make sense?
            last = render; // predict second field from the first
        render->info.mpeg.forward_reference      = last->surface;
    }

    ff_vdpau_add_data_chunk(s, buf, buf_size);

    render->info.mpeg.slice_count                = slice_count;

    if (slice_count)
        ff_draw_horiz_band(s, 0, s->avctx->height);
    render->bitstream_buffers_used               = 0;
}

/* @}*/
Commit	Line	Data
369122dd NC	1	/*
	2	* Video Decode and Presentation API for UNIX (VDPAU) is used for
	3	* HW decode acceleration for MPEG-1/2, H.264 and VC-1.
	4	*
406792e7	5	* Copyright (c) 2008 NVIDIA
369122dd NC	6	*
	7	* This file is part of FFmpeg.
	8	*
	9	* FFmpeg is free software; you can redistribute it and/or
	10	* modify it under the terms of the GNU Lesser General Public
	11	* License as published by the Free Software Foundation; either
	12	* version 2.1 of the License, or (at your option) any later version.
	13	*
	14	* FFmpeg is distributed in the hope that it will be useful,
	15	* but WITHOUT ANY WARRANTY; without even the implied warranty of
	16	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
	17	* Lesser General Public License for more details.
	18	*
	19	* You should have received a copy of the GNU Lesser General Public
	20	* License along with FFmpeg; if not, write to the Free Software
	21	* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
	22	*/
	23
	24	#include <limits.h>
	25	#include "avcodec.h"
	26	#include "h264.h"
	27
	28	#undef NDEBUG
	29	#include <assert.h>
	30
8423186e	31	#include "vdpau.h"
369122dd NC	32	#include "vdpau_internal.h"
	33
	34	/**
	35	* \addtogroup VDPAU_Decoding
	36	*
	37	* @{
	38	*/
	39
17170313	40	void ff_vdpau_h264_set_reference_frames(MpegEncContext *s)
369122dd	41	{
17170313	42	H264Context *h = s->avctx->priv_data;
369122dd NC	43	struct vdpau_render_state * render, * render_ref;
	44	VdpReferenceFrameH264 * rf, * rf2;
	45	Picture * pic;
	46	int i, list, pic_frame_idx;
	47
	48	render = (struct vdpau_render_state*)s->current_picture_ptr->data[0];
	49	assert(render);
	50
	51	rf = &render->info.h264.referenceFrames[0];
	52	#define H264_RF_COUNT FF_ARRAY_ELEMS(render->info.h264.referenceFrames)
	53
	54	for (list = 0; list < 2; ++list) {
	55	Picture **lp = list ? h->long_ref : h->short_ref;
	56	int ls = list ? h->long_ref_count : h->short_ref_count;
	57
	58	for (i = 0; i < ls; ++i) {
	59	pic = lp[i];
	60	if (!pic \|\| !pic->reference)
	61	continue;
	62	pic_frame_idx = pic->long_ref ? pic->pic_id : pic->frame_num;
	63
	64	render_ref = (struct vdpau_render_state*)pic->data[0];
	65	assert(render_ref);
	66
	67	rf2 = &render->info.h264.referenceFrames[0];
	68	while (rf2 != rf) {
	69	if (
	70	(rf2->surface == render_ref->surface)
	71	&& (rf2->is_long_term == pic->long_ref)
	72	&& (rf2->frame_idx == pic_frame_idx)
	73	)
	74	break;
	75	++rf2;
	76	}
	77	if (rf2 != rf) {
	78	rf2->top_is_reference \|= (pic->reference & PICT_TOP_FIELD) ? VDP_TRUE : VDP_FALSE;
	79	rf2->bottom_is_reference \|= (pic->reference & PICT_BOTTOM_FIELD) ? VDP_TRUE : VDP_FALSE;
	80	continue;
	81	}
	82
	83	if (rf >= &render->info.h264.referenceFrames[H264_RF_COUNT])
	84	continue;
	85
	86	rf->surface = render_ref->surface;
	87	rf->is_long_term = pic->long_ref;
	88	rf->top_is_reference = (pic->reference & PICT_TOP_FIELD) ? VDP_TRUE : VDP_FALSE;
	89	rf->bottom_is_reference = (pic->reference & PICT_BOTTOM_FIELD) ? VDP_TRUE : VDP_FALSE;
	90	rf->field_order_cnt[0] = pic->field_poc[0];
	91	rf->field_order_cnt[1] = pic->field_poc[1];
	92	rf->frame_idx = pic_frame_idx;
	93
	94	++rf;
	95	}
	96	}
	97
	98	for (; rf < &render->info.h264.referenceFrames[H264_RF_COUNT]; ++rf) {
	99	rf->surface = VDP_INVALID_HANDLE;
	100	rf->is_long_term = 0;
	101	rf->top_is_reference = 0;
	102	rf->bottom_is_reference = 0;
	103	rf->field_order_cnt[0] = 0;
	104	rf->field_order_cnt[1] = 0;
	105	rf->frame_idx = 0;
	106	}
107	}
108
c639fc72 CEH	109	void ff_vdpau_add_data_chunk(MpegEncContext *s,
c639fc72 CEH	110	const uint8_t *buf, int buf_size)
369122dd	111	{
369122dd NC	112	struct vdpau_render_state * render;
	113
	114	render = (struct vdpau_render_state*)s->current_picture_ptr->data[0];
	115	assert(render);
	116
6cc01c24 CEH	117	render->bitstream_buffers= av_fast_realloc(
	118	render->bitstream_buffers,
	119	&render->bitstream_buffers_allocated,
	120	sizeof(render->bitstream_buffers)(render->bitstream_buffers_used + 1)
369122dd NC	121	);
369122dd NC	122
6cc01c24 CEH	123	render->bitstream_buffers[render->bitstream_buffers_used].struct_version = VDP_BITSTREAM_BUFFER_VERSION;
	124	render->bitstream_buffers[render->bitstream_buffers_used].bitstream = buf;
	125	render->bitstream_buffers[render->bitstream_buffers_used].bitstream_bytes = buf_size;
	126	render->bitstream_buffers_used++;
369122dd NC	127	}
369122dd NC	128
17170313	129	void ff_vdpau_h264_picture_complete(MpegEncContext *s)
369122dd	130	{
17170313	131	H264Context *h = s->avctx->priv_data;
369122dd NC	132	struct vdpau_render_state * render;
	133
	134	render = (struct vdpau_render_state*)s->current_picture_ptr->data[0];
	135	assert(render);
	136
	137	render->info.h264.slice_count = h->slice_num;
	138	if (render->info.h264.slice_count < 1)
	139	return;
	140
	141	for (int i = 0; i < 2; ++i) {
	142	int foc = s->current_picture_ptr->field_poc[i];
	143	if (foc == INT_MAX)
	144	foc = 0;
	145	render->info.h264.field_order_cnt[i] = foc;
	146	}
	147
	148	render->info.h264.is_reference = s->current_picture_ptr->reference ? VDP_TRUE : VDP_FALSE;
	149	render->info.h264.frame_num = h->frame_num;
	150	render->info.h264.field_pic_flag = s->picture_structure != PICT_FRAME;
	151	render->info.h264.bottom_field_flag = s->picture_structure == PICT_BOTTOM_FIELD;
	152	render->info.h264.num_ref_frames = h->sps.ref_frame_count;
	153	render->info.h264.mb_adaptive_frame_field_flag = h->sps.mb_aff;
	154	render->info.h264.constrained_intra_pred_flag = h->pps.constrained_intra_pred;
	155	render->info.h264.weighted_pred_flag = h->pps.weighted_pred;
	156	render->info.h264.weighted_bipred_idc = h->pps.weighted_bipred_idc;
	157	render->info.h264.frame_mbs_only_flag = h->sps.frame_mbs_only_flag;
	158	render->info.h264.transform_8x8_mode_flag = h->pps.transform_8x8_mode;
	159	render->info.h264.chroma_qp_index_offset = h->pps.chroma_qp_index_offset[0];
	160	render->info.h264.second_chroma_qp_index_offset = h->pps.chroma_qp_index_offset[1];
	161	render->info.h264.pic_init_qp_minus26 = h->pps.init_qp - 26;
	162	render->info.h264.num_ref_idx_l0_active_minus1 = h->pps.ref_count[0] - 1;
	163	render->info.h264.num_ref_idx_l1_active_minus1 = h->pps.ref_count[1] - 1;
	164	render->info.h264.log2_max_frame_num_minus4 = h->sps.log2_max_frame_num - 4;
	165	render->info.h264.pic_order_cnt_type = h->sps.poc_type;
	166	render->info.h264.log2_max_pic_order_cnt_lsb_minus4 = h->sps.log2_max_poc_lsb - 4;
	167	render->info.h264.delta_pic_order_always_zero_flag = h->sps.delta_pic_order_always_zero_flag;
	168	render->info.h264.direct_8x8_inference_flag = h->sps.direct_8x8_inference_flag;
	169	render->info.h264.entropy_coding_mode_flag = h->pps.cabac;
	170	render->info.h264.pic_order_present_flag = h->pps.pic_order_present;
	171	render->info.h264.deblocking_filter_control_present_flag = h->pps.deblocking_filter_parameters_present;
	172	render->info.h264.redundant_pic_cnt_present_flag = h->pps.redundant_pic_cnt_present;
	173	memcpy(render->info.h264.scaling_lists_4x4, h->pps.scaling_matrix4, sizeof(render->info.h264.scaling_lists_4x4));
	174	memcpy(render->info.h264.scaling_lists_8x8, h->pps.scaling_matrix8, sizeof(render->info.h264.scaling_lists_8x8));
	175
	176	ff_draw_horiz_band(s, 0, s->avctx->height);
6cc01c24	177	render->bitstream_buffers_used = 0;
369122dd NC	178	}
369122dd NC	179
d37edddc NC	180	void ff_vdpau_mpeg_picture_complete(MpegEncContext s, const uint8_t buf,
	181	int buf_size, int slice_count)
	182	{
	183	struct vdpau_render_state * render, * last, * next;
	184	int i;
	185
	186	render = (struct vdpau_render_state*)s->current_picture_ptr->data[0];
	187	assert(render);
	188
	189	/* fill VdpPictureInfoMPEG1Or2 struct */
	190	render->info.mpeg.picture_structure = s->picture_structure;
	191	render->info.mpeg.picture_coding_type = s->pict_type;
	192	render->info.mpeg.intra_dc_precision = s->intra_dc_precision;
	193	render->info.mpeg.frame_pred_frame_dct = s->frame_pred_frame_dct;
	194	render->info.mpeg.concealment_motion_vectors = s->concealment_motion_vectors;
	195	render->info.mpeg.intra_vlc_format = s->intra_vlc_format;
	196	render->info.mpeg.alternate_scan = s->alternate_scan;
	197	render->info.mpeg.q_scale_type = s->q_scale_type;
	198	render->info.mpeg.top_field_first = s->top_field_first;
	199	render->info.mpeg.full_pel_forward_vector = s->full_pel[0]; // MPEG-1 only. Set 0 for MPEG-2
	200	render->info.mpeg.full_pel_backward_vector = s->full_pel[1]; // MPEG-1 only. Set 0 for MPEG-2
	201	render->info.mpeg.f_code[0][0] = s->mpeg_f_code[0][0]; // For MPEG-1 fill both horiz. & vert.
	202	render->info.mpeg.f_code[0][1] = s->mpeg_f_code[0][1];
	203	render->info.mpeg.f_code[1][0] = s->mpeg_f_code[1][0];
	204	render->info.mpeg.f_code[1][1] = s->mpeg_f_code[1][1];
	205	for (i = 0; i < 64; ++i) {
	206	render->info.mpeg.intra_quantizer_matrix[i] = s->intra_matrix[i];
	207	render->info.mpeg.non_intra_quantizer_matrix[i] = s->inter_matrix[i];
	208	}
	209
	210	render->info.mpeg.forward_reference = VDP_INVALID_HANDLE;
	211	render->info.mpeg.backward_reference = VDP_INVALID_HANDLE;
	212
	213	switch(s->pict_type){
	214	case FF_B_TYPE:
	215	next = (struct vdpau_render_state*)s->next_picture.data[0];
	216	assert(next);
	217	render->info.mpeg.backward_reference = next->surface;
	218	// no return here, going to set forward prediction
	219	case FF_P_TYPE:
	220	last = (struct vdpau_render_state*)s->last_picture.data[0];
	221	if (!last) // FIXME: Does this test make sense?
	222	last = render; // predict second field from the first
	223	render->info.mpeg.forward_reference = last->surface;
	224	}
	225
	226	ff_vdpau_add_data_chunk(s, buf, buf_size);
	227
	228	render->info.mpeg.slice_count = slice_count;
	229
	230	if (slice_count)
	231	ff_draw_horiz_band(s, 0, s->avctx->height);
	232	render->bitstream_buffers_used = 0;
	233	}
	234
369122dd	235	/* @}*/